Scientists may be reluctant to share their data until they can publish their findings, at which time they _have_ to pony up something substantial to get it (their article) by reviewers. That could be similar to an agency reluctant to share their geospatial data when they are not real confident in it and realize that announcing it via metadata may bring to them more attention or scrutiny of their data than they want. As far as reluctance to sharing data goes, scientists are probably no worse in this regard than GISers (and many GISers _are_ scientists).
In regards to the question about science-based groups implementing metadata: Other sciences were documenting their data (without calling it metadata) long before GIS became an acronym. GIS is really a Johnny-come-lately to metadata.
In many respects, building a GIS data set from various sources then documenting it with metadata is much like synthesizing a chemical compound from various reagents, then describing the procedure. A journal article describing such a synthesis would typically have a title, authors, keywords, an abstract, an introduction which would encapsulate the purpose or the use for the compound, a list of reagents and apparatus, the reaction procedure and conditions, a characterization of the compound (melting point, boiling point, smell, chemical structute, infrared or NMR spectra), a conclusion, and some references. That is a lot like the elements of the CSDGM in a different data environment. One need only pick up a copy of the Merck Index to get highly distilled (!) metadata on thousands of compounds or perform an organic chemical structural search using the Beilstein database to get an appreciation for the value of a _lot_ of metadata.
I am sure others can provide examples from other sciences. I would bank on astronomy having a long tradition for metadata.
From Phoenix, AZ
Why haven`t GIS software vendors been more proactive in
the development of metadata collection tools?
Hmmm. Well, I have never asked a vendor why they haven't done this, so I can only speculate. Get your boots on!
I suspect that the complexity of SDTS, and the limited lack of acceptance and use of that data exchange standard made vendors somewhat leery of another complex federal standard which also might not see wide acceptance. Perhaps they see their strength in GIS data processing algorithms, but not in what is for the most part a text processing task. Until users start jumping ship on a GIS with good data processing capabilities for another GIS with metadata capabilities, vendors will probably not consider it an essential capability to build into their tools.
I think if an exchange standard such as the one outlined at: http://fgdc.er.usgs.gov/Communications/Metadata/encoding797.html were adopted, then GIS software companies and metadata tool developers both would feel more comfortable with the CSDGM, and they would have a target for both input and output of metadata.
From Dave Murray at the University of Colorado at Denver
Metadata can point out glaring errors in your dataset. How do
you resolve the personal problems that arise out of
metadata creation?
If the dataset producing agency is serious about the modeling that is done with these data and if metadata points to glaring errors in them, then the agency should rightly be concerned about it and learn from the experience. If an agency does not have a metadata policy and data quality control procedures in place which are fully endorsed by management then this exposure should never cause personnel problems at the GIS Analyst or Technician level; it should precipitate a process leading to responsibility for metadata and quality control being assigned to particular individuals and the time and resources necessary to perform these tasks allocated as a standard part of any project.
Name and location not given
Do you feel that records that are complete, but do not pass the
FGDC "metadata parser" are acceptable, or at least
useful.
If by 'complete' you mean CSDGM compliant, then I would certainly say the metadata was useful. As far as using passing the metadata parser as a measure of acceptance goes, well, it is a useful (very useful) benchmark, but it has no official sanction, and fully CSDGM compliant metadata may fail miserably to pass through mp. Whether or not the metadata is 'acceptable' depends a lot on the relationship between you, the metadata producer and me, the metadata acceptor.
The first party to satisfy is yourself. Metadata will nowhere be more useful than it is internally to the organization that produced it. If it serves the purpose of preserving the value and usefulness of the data within _your_ organization, then, by golly, it passes the most important measure of acceptance.
If I were getting the metadata along with a data set from you that I intended to use for GIS analysis, then I would be overjoyed to get anything like complete metadata. Usually the metadata I get is a filename, and I have to make many follow-up phone calls to get the metadata I need to use the data effectively and confidently. Of course, if the metadata you gave me would pass mp, then in would be easy for me to incorporate it with our existing metadata, and to suck it into xtme so I could easily modify it to reflect any processing our organization did on it.
If I were running a Clearinghouse for geospatial data, then my acceptance criteria could be more narrow. I will admit, that when I helped establish Wisconsin's NSDI Clearinghouse I was desperate for metadata and would accept almost anything. I was willing to (and did) spend hours editing the metadata so it could pass mp and be indexed correctly for a Clearinghouse search. Now that I'm old and crotchety though I would be much more insistent (in the Clearinghouse situation) that 'acceptable' metadata would have to pass mp. There are several reasons for this. The first is that after some editing and filtering to get icky metadata into mp, I may find that the metadata is non-compliant - it is missing a section or otherwise structurally broken. This will result in a flurry of phone calls, and unless I am willing to do the repair myself, it will be a couple weeks (or never) before I get the corrected metadata back from the original submitter. If metadata submitters ensure their metadata is compliant with the metadata parser _before_ they submit metadata to a Clearinghouse, then an entire round of headache can be eliminated for the Clearinghouse operator and the metadata submitter. The second is, that by accepting metadata which will not pass mp, I have set myself up to always have to filter or edit metadata from some submitters, which is a time committment. It makes sense for the metadata submitter to standardize the procedure to convert their metadata to a form mp can accept, and to validate it, before submitting it to a Clearinghouse. Ultimately the metadata has to pass mp (mostly) in order to be converted to SGML and be available for a Clearinghouse metadata search.
From Lucille Bish in Ontario, Canada
We are developing a metadata application for municipal use. Do
you know of a thesaurus of key words related to
municipal spatial data?
None that I am aware of. In fact there are no good thesauri for many other applications of GIS and GIS themes. This is an issue which has been raised many times, but no champion has risen to tackle the task. A thesaurus wouldn't just be useful so documentors would know words to use to document their data, it would also make searches of Clearinghouses more effective.
There are some useful keywords available from the Global Change Master Directory (http://gcmd.gsfc.nasa.gov/cgi-bin/mduser_dir/earth_sci_keywords) , but certainly nothing comprehensive that would give us a vocabulary across the many applications of GIS in natural resources, geodetic control, transportation, utilities, etc. Building a 'proper' thesaurus is no mean task, and it has been suggested that if FGDC were to support development of such an item through a CCAP grant say, it must surely involve a library or librarian familiar with such resources. Less complex than a thesaurus is a 'controlled vocabulary.' That is also not available, to my knowledge.
I and many others are interested in this keyword subject, and the potential that we might get practitioners in specific areas to suggest keywords cascaded in specificity from general to more specific is very appealing. If you do develop such a list, then I would be pleased to learn of it, as would, I am sure, many other on the nsdi-l list involved in metadata production.
From Kevin Kane at Iowa State University
1. What is "Metadata Lite?"
It is an non-official term for 'just enough metadata to get by.' At the minimum it might consist of the mandatory elements in sections 1 and 7 of the CSDGM. It might also include those elements and a few more which some would like to call the 'core' elements of the CSDGM, as if those were the 'most important' parts. Others reject the idea that the standard can be picked apart so easily. To suggest or to endorse that a subset of the CSDGM is really the important part would encourage, even validate to some extent, the production of incomplete metadata as a matter of course.
In practice, most users are not bound by any federal mandates to create CSDGM compliant metadata, so they can create metadata at any level of detail they need for their operations. This obviates the need the need to define any feel-good thresholds like 'Metadata Lite.'
A: I can think of two reasons. First, having worked with the Federal content standard as well as a "local" implementation using DBF files, I think there are some complex software implementation issues. The vendors are smarter than I am (I hope), but I suspect they have these problems too. Second, I think they want to hear from YOU. They will build it if they know you want it. Won't they?
Name and location not given
Do you feel that records that are complete, but do not pass
the FGDC "metadata parser" are acceptable, or at least
useful.
A: ABSOLUTELY. The parser will yield many warning and error conditions even when required fields are filled. You may just have filled them in a format or with content that the parser is not expecting. I see no easy answer, except to say that we'll be evaluating all error messages and judging what if anything we can do to address them.
From Lorri Peltz-Lewis
Can you address how to change an office that lacks
commitment to one that has a commitment? Even with an
Executive Order, some Federal offices still don't have a
commitment; it's seen as an unfunded mandate.
A: All the GIS literature is clear. Sometimes the critical factor can be a "champion" or a crisis or a funding opportunity. Maybe you can make those things happen. Maybe you can just continue to do "the right thing" to the extent you can, without management support.
From Jonathon Levine at the University of Michigan
What approaches exist to ensure that metadata "sticks" with
its data set as the data set travels from hand to hand?
A: An approach we've taken is to include bunches of metadata into DBF files which are embedded within the coverages. However, that's software dependent. A technical technique you can use is to "ZIP" your data layers for delivery to other parties, and to ALWAYS ZIP various "readme" or metadata or other files into the same ZIPfile.
From Raj Aggarwala at the University of Michigan
1. To be acceptable metadata, who is responsible for
preparing meta data sets - this is important as with
respect to the "POSSIBLE RELIABILITY OF THE METADATA
ITSELF" if prepared by personnel having only cursory
knowledge of the data sets.
A: I believe that quality metadata can best be created with some division of labor. That is, you might have a technician who can perform a lot of the routine metadata creation and editing. AND a database manager who assures some level of completeness and commonality across metadata sets. AND the "data manager" responsible for the particular data layer, who is the person who knows the most about the usefulness and development of the data layer. I would say that persons with only cursory knowledge can create metadata files which are cursory; i.e. only good as basic pointers.
2. Shouldn't it be mandatory that ONLY DATA PRODUCERS provide metadata to be of the highest reliability. Or develop a reliability index of Metadata itself.
A: I think this raises hard questions about what kinds of organizations are in a position to issue mandates. FGDC discussion of "area integrators" and other roles within the NSDI have touched on these issues, with no resolution of which I am aware.
From Columbus
Rather than inventory old data, why not begin by capturing
metadata on new data creation? Why worry about old data?
A: Some of our "best" data is "old" data. If that's the data that people want and use, it seems to me that you'd want to make the effort to get them good documentation to go along with it. But you might have different meanings for "old." "Old" in the sense of passing time is different than "old" in the sense of out-of-date.
Bruce Westcott, Executive Director
VT Center for Geographic Information, Inc.
206 Morrill Hall
Burlington, VT 05405-0106
(802)656-4277 FAX 656-0776
http://geo-vt.uvm.edu
last updated by David Hart on November 12, 1997